English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Black-box query-based attacks constitute significant threats to MachineLearning as a Service (MLaaS) systems since they can generate adversarialexamples without accessing the target model's architecture and parameters.Traditional defense mechanisms, such as adversarial training, gradient masking,and input transformations, either impose substantial computational costs orcompromise the test accuracy of non-adversarial inputs. To address thesechallenges, we propose an efficient defense mechanism, PuriDefense, thatemploys random patch-wise purifications with an ensemble of lightweightpurification models at a low level of inference cost. These models leverage thelocal implicit function and rebuild the natural image manifold. Our theoreticalanalysis suggests that this approach slows down the convergence of query-basedattacks by incorporating randomness into purifications. Extensive experimentson CIFAR-10 and ImageNet validate the effectiveness of our proposedpurifier-based defense mechanism, demonstrating significant improvements inrobustness against query-based attacks.

PuriDefense: Randomized Local Implicit Adversarial Purification for Defending Black-box Query-based Attacks